Texture analysis of lung nodules in computerized tomography images using functional diversity

Although lung cancer is one of the leading causes of cancer deaths worldwide, the chances of survival are higher in the early stages. One of the best tools for diagnosis is computerized tomography. The main problem with this method is that it depends directly on the specialist who is analyzing the image, since the process involved is tiring, and can lead to error. Computer-aided detection systems have emerged as a way to help these specialists. This work presents the use of descriptors based on functional diversity indexes to reduce the number of false positives. Our method can reach an accuracy of 97.73%, a sensitivity of 98.4%, Kappa index of 0.941, and a number of false positives per scan of up to three. Based on the results obtained, the use of a functional diversity index is shown to be a robust method that can be used in a real CAD system. (c) 2020 Elsevier Ltd. All rights reserved.